Other BSODs at startup mystery - tl;dr + daily log

tl;dr - get BSOD sometimes during startup, it's not Windows, drivers or drives. Happens even with RAM A-XMP profiles and Try It! Any usual cause? Or lost cause and just live with it?

Full version + log - When I upgraded to new motherboard (Mortar Max), CPU (Ryzen 3600) and RAM (2 x 8GB Corsair Vengeance LPX @ 3200 MHz), I started getting BSODs during startup (just after splash, before desktop) but it restarts itself so not a big deal. I figured it was due to driver conflict, so I got an SSD anyway and used that as a boot drive. But nope, still happening even with old drive disconnected.

My next check was the RAM. I haven't done MemTest but I've done Windows' memory test with default settings and it found nothing. I reinserted the RAM and have tried both A-XMP profiles (which seem to be identical anyway) and Memory Try It! but still get them. Only thing left to try is manually setting RAM speed to the 3200 MHz it should be at (it defaults at 2133) and MemTest. Maybe update BIOS if there is one.

I even tried using an NVIDIA driver from before the BSODs.

PSU? Guessing not.
Motherboard? Sure hope not.
GPU memory being daft? Random guess.

I also got a couple of new things now, as seen in the log. Also, one more random oddity: Using VLC, now when I pause I risk the video getting garbled, which can be undone by clicking another time in the video.

Now the log. I've had this since June but only got the SSD recently, so I started logging a few weeks ago. Enjoy:

I think I started with A-XMP profile 2. I boot up every day.

4th oct - BSOD

6th - BSOD

7th - system thread exception not handled BSOD

8th - uninstall nvidia driver and install new driver

9 - system thread exception not handled BSOD
9 - HDD disconnected
9 - multiple restarts, no bsod

10 - memory_management BSOD but after boot

12 - HDD reconnected

14 - memory_management long after boot
14 - microsoft memory test clear, uninstall nvidia driver and reinstall March nvidia driver
14 - No BSOD but shuts down a few secs into boot
14 - HDD disconnected

15 - kernel_security_check_failure BSOD
15 - Reconnected hdd, reinserted RAM, change to a-xmp 1

20 - system_service_exception BSOD

21 - No BSOD but restarts like one again
21 - Change a-xmp to 2

22 - Black and hear a windows notification or 'device inserted' noise, after a couple minutes press power button (which usually does nothing but will turn off PC if it's screwing up) and screen goes orange and cursor can be moved. Another minute or 2, I turn off (holding power button)

22 - Boots with .dll error (realtek), then try open Firefox and supposedly .dll missing and XCOM missing. Jsut restart myself.

24 - kernel_security_check_failure BSOD
24 - turn off a-xmp altogether and use Memory try it!

25 - kernel BSOD

 

I got a new BSOD: video scheduler internal error

BSODs are a mystery never to be solved, but the word 'video' obviously jumps out. The only other video card I have is an 8800 GT, but there's no point getting that dusty thing out to check it's the card when I'd just put it back in anyway.

I've just played all of Watch Dogs Legion fine, it's still only coming up during boot. Even on days it doesn't happen, something funny like not being able to search in File Explorer happens and I have to restart anyway to fix it. But with an SSD now it's still not a big problem.

UPDATE EDIT

'IRQL not less or equal'

Another new one. 2 days in a row!

 

UPDATE:

I luckily discovered a program called BlueScreenView that gives more info on BSOD. It's only showing 5, but 3 were caused by driver 'ntoskrnl.exe'.

 

Set your RAM to rated speed using the XMP setting, then download this, set up a USB stick and boot from it. The Windows in built memory test is completely useless, and everything you're listing sounds like RAM problems.

 

It's always been at the speed it should be, I've tried both (seemingly same) XMP profiles as above, and already done memtest64 and 86 with no errors.

 

Did you run Memtest86 in UEFI mode?
Have you tried reseating your CPU?

 

I don't know what that means but I put it on a USB stick and booted from it.

I don't recall if I got the BSOD before reseating the CPU but I didn't have any issues doing it (hey it's hard to get wrong) and temps are fine.

What I will try in future is reseating the GPU, chkdisk, using the unadjusted (but incorrect) RAM speed of 2133MHz and using the other 2 RAM slots instead.

 

Try 1 stick of ram, move slots, try the other stick on its own too. test the ram. Try a fresh os install on a spare drive if this is an existing windows install. Reseat cpu. test the CPU there must be some amd cpu software to test it with?

 

My bad, when choosing to boot from the USB stick, your BIOS should give your standard options (BIOS/MBR) or modern options (UEFI/GPT). Memtest86 stopped development on the MBR side, so the newer UEFI booted mode is far superior in finding faults and doesn't require you to run it for an arbitrary length of time (it runs 4 passes of all tests).

I agree with the majority of this, except for "some amd cpu software to test it with". You can stress test but as far as I know, there isn't any manufacturer specific CPU test software for AMD or Intel.

 

UPDATE: Updating the BIOS didn't seem to work (I got a BSOD but quite a while after booting), but keeping the RAM at default 2133MHz (rather than the 3200Mhz it should be on) hasn't got one in the past few days yet.

It didn't give me any option, just went straight into it. It was 4 tests but I only waited around for 2.

 

if you want to try some different RAM, shout out, I'll be able to lend you some for testing.

 

Thanks mang. Still no BSOD from having it at 2133MHz so if I could try RMAing the RAM without losing use of my PC for however long that takes, that'd be great!

 

PM'ed

 

Had the new RAM sticks in for a week now and no BSODs, so it must be the RAM.

Now I can look into what this RMAing is all about.

 

I do like a problem being solved.

 

It is thanks to spolsh!

 

Good old @spolsh ,his trumps coming up again

 

I put in an RMA request and they said it's not in the QVL list. I've replied saying it's in their Ryzen 3000 compatibility list (https://help.corsair.com/hc/en-us/articles/360030465392-AMD-RYZEN-2000-3000-SERIES-Memory-Compatibility-List-) and is also a speed compatible with the motherboard (https://www.msi.com/Motherboard/B450M-MORTAR-MAX/Specification), and just generally it's the most popular pair of sticks everyone's buying to go with their new 3600 and is considered compatible by all sorts of other sites.

But I don't know how far that will get me. I wasn't aware of this secret list. Not that I really needed to, but I did check motherboard RAM compatibility before buying. I didn't know the secret code known as 'QVL'. Then again, the entire brand of the RAM I'm using now (TeamGroup) isn't even on the QVL list at all!

 

QVL lists are [censored]. Basically it's just "did the manufacturer bother to test?" As the sheer variety of different brands, types, timings etc. out there are impossible to realistically test, I view QVL lists as little more than a "get out of jail free" card for companies who can't be bothered to support their customers.

The only exception to this for me are server-grade boards and ECC-Reg RAM, where I will always try to get stuff on the QVL list.

To further exemplify why I think QVL lists are a bit of a con; I've been told before that a GTX 1080Ti (which was, at the time, the second best consumer card available) was "not supported" on an X99 board. Uh...

As for RAM, sometimes it's a crapshoot. I've had different revisions of the same RAM behave differently on the same board - but most of the time, if timings/speed/voltage match, I've not had any trouble, even mixing kits on triple channel boards.

RAM outside of JEDEC specs are another one. For example, one set RAM was sold as 3333MHz RAM, but is only "officially" supported at 2666MHz (anything else is "overclocking"). Uh...

 

Corsair asked me to test each stick individually in the slot furthest from the CPU with XMP on. As soon as I replied saying I've tried 1 stick so far and got a BSOD and now trying the other, they instantly approved an RMA.

I'm discovering more and finding out this is common. Posts are a few years old, but there's loads resorting to lower speeds to get this RAM stable. I did use the Ryzen calculator and Thaiphoon a while ago and the supposed safest timings didn't even POST. Maybe it's a stick, or just a scam, or Ryzen being incompatible.

I'll RMA them at some point and whatever I get back, I'll see how it goes or just sell them and buy the RAM I'm using now off Ebay etc., or from spolsh if they fancy, because so far I've had no problems and they're better timings too.